Fix dead_code_pub_in_binary false positive on library crates

[Kokoro2336]

Fixes #159078

Fix dead_code_pub_in_binary false positive on public items when compiling library crates with --all-targets flag

[rustbot]

r? @jackh726

rustbot has assigned @jackh726.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: compiler
• compiler expanded to 75 candidates
• Random selection from 19 candidates

[fmease]

Wasn't #74970 (comment) a somewhat important use case for this lint firing under --test? IINM that would no longer be served. cc @RalfJung

After all, crates under tests are binaries with an auto-generated main function. However, this might be considered an implementation detail, so I understand the desire to change this.

Ideally, we would have some replacement for the linked use case, then.

Disclaimer: I've only skimmed everything so I might've missed something obvious.

[RalfJung]

That test I mentioned there was a ui test, i.e. a regular binary crate -- no --test involved.

[Kokoro2336]

All the crates compiled with a --test flag would be considered as CrateType::Executable, making it tricky to recognize the actual type of the compiled crate when running check_mod_deadness. I used opts.crate_types to distinguish it, though I don't think it's an optimal choice since it's read from cmd args.

[Kokoro2336]

Plus, I've supplied comparison tests between normal crate and library crate.